Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 18 Jan 1999 23:30:33 -0800 (PST)
From:      Julian Elischer <julian@whistle.com>
To:        Matthew Dillon <dillon@apollo.backplane.com>
Cc:        Archie Cobbs <archie@whistle.com>, hackers@FreeBSD.ORG
Subject:   Re: Found problem w/ Paging performance over NFS
Message-ID:  <Pine.BSF.3.95.990118233009.9336C-100000@current1.whistle.com>
In-Reply-To: <199901190709.XAA90152@apollo.backplane.com>

next in thread | previous in thread | raw e-mail | index | archive | help
please commit luoqi's NFS fix....


On Mon, 18 Jan 1999, Matthew Dillon wrote:

> 
> :Matthew Dillon writes:
> :>     Ok.  I just figured out the interaction -- in fact, we *can't* test m->busy
> :>     in this section of code because of the following sequence:
> :> 
> :> 	vm_fault -> vnode_pager_getpages -> ffs_getpages -> ffs_read -> 
> :> 	    cluster_read -> getblk -> allocbuf ( lockup )
> :> 
> :>     ffs_getpages converts the PG_BUSY into a vm_page_t->busy before getting into
> :>     the cluster stuff, and thus allocbuf() deep down cannot sleep on
> :>     vm_page_t->busy.
> :> 
> :...
> :Yow. I don't think it's possible to overcomment the VM subsystem.. :-)
> :
> :-Archie
> 
>     Heh.  Believe you me, I'm commenting everything that takes me more then
>     5 minutes to puzzle out, and then some!
> 
>     This stuff is starting to look pretty good, I'll open the patchset up
>     to anyone who is interested in testing it.  You can pop it off my site.
> 
>     WARNING!  Make backups of your original files or be prepared to re-check
>     them out fresh from the cvs tree.  This is highly experimental code.
> 
> 		(for latest)
> 
> 	http://www.backplane.com/FreeBSD/
> 
> 		(last snapshot, fresh at posting time)
> 
> 	http://www.backplane.com/FreeBSD/dillon-swapper-A15.tgz
> 	http://www.backplane.com/FreeBSD/FreeBSDVMWork.html
> 
>     I've been testing the stuff on a 48MB diskless workstation.  I've 
>     successfully run a -j14 buildworld and it's running a -j18 right now.
>     If that works, I'll try a -j30 next.  It's actually kinda hard to get
>     it to grind on a 100BaseTX LAN with the NFS paging fixes in place :-)
> 
>     BTW, the patch set also includes Luoqi's NFS fix.
> 
>     					-Matt
> 
> 					Matthew Dillon 
> 					<dillon@backplane.com>
> 
> 
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-hackers" in the body of the message
> 


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.95.990118233009.9336C-100000>